Gaussian function
PulseAugur coverage of Gaussian function — every cluster mentioning Gaussian function across labs, papers, and developer communities, ranked by signal.
11 day(s) with sentiment data
-
New theory explores scaling laws in contrastive representation learning
Researchers have developed a theoretical framework for understanding scaling laws in contrastive representation learning. The paper analyzes a sketched linear model under a paired Gaussian latent-variable setup, derivin…
-
New Laplace--Fisher Gate Identity Enhances Score Estimation in Bayesian Inverse Problems
Researchers have developed a new method called the Laplace--Fisher Gate Identity (LFGI) for estimating scores in sampling from unnormalized targets. This method uses matrix-valued blending coefficients, or gates, to opt…
-
New paper models belief formation geometry under noisy observation
A new arXiv paper explores the geometric costs associated with belief formation in finite systems that operate with noisy observations. The research models the process as optimal transport in Wasserstein space, reweight…
-
New randomization method enhances black-box AI stability
Researchers have developed a new methodology for stabilizing black-box algorithms, which are increasingly crucial for trustworthy AI. This task-oriented randomization approach adapts to diverse input data, including com…
-
New research offers advanced methods for image denoising
Two new research papers propose novel methods for image denoising. The first paper introduces a Mixed-norm TV (MixTV) model that aims to reduce noise while preserving image edges, demonstrating improved effectiveness ov…
-
New framework for distributional Granger causality developed
A new framework for distributional Granger causality has been developed, extending beyond the conditional mean to analyze predictive dependence in time series. This approach considers distributional features like scale,…
-
New theory links Mahalanobis Cosine Similarity to probe performance
Researchers have theoretically and empirically demonstrated that Mahalanobis Cosine Similarity (MCS) is a strong predictor of a linear probe's Out-of-Distribution AUROC. This relationship holds across various models, la…
-
New zero-inflated Gaussian distributions boost sparse optimization algorithms
Researchers have developed a novel approach to enhance estimation-of-distribution algorithms (EDAs) for optimization problems with sparse parameter spaces. By employing multivariate zero-inflated Gaussian (ZIG) distribu…
-
New PC-TGS framework enhances wireless channel prediction using LiDAR and radio data
Researchers have developed a new framework called Point-Cloud-Assisted Tangent Gaussian Splatting (PC-TGS) to improve channel prediction in wireless networks. This method integrates sparse radio measurements with dense …
-
InfoNCE objective induces Gaussian distribution in AI representations
Researchers have demonstrated that the InfoNCE contrastive learning objective inherently promotes a Gaussian distribution within learned representations. This finding was established through theoretical analysis under s…
-
GauS framework optimizes operator scheduling using Gaussian reparameterization
Researchers have introduced GauS, a novel differentiable framework for optimizing operator scheduling in software compilation and hardware synthesis. Unlike previous methods that used categorical distributions, GauS emp…
-
New method improves AI's ability to solve inverse problems
Researchers have developed a new method called Exact Posterior Score (EPS) for solving linear inverse problems using diffusion and flow-based models. This technique derives the exact posterior score in closed form for l…
-
New method for multivariate time series prediction sets unveiled
Researchers have introduced filtered conformal ellipsoids, a novel method for joint prediction sets in multivariate time series. This approach utilizes a state-space filter to emit predictive means and covariances, whic…
-
New method characterizes recovery thresholds for hidden weighted sparse graphs
Researchers have developed a unified characterization for the information-theoretic limits of recovering hidden structures within noisy, high-dimensional data. The study focuses on identifying an unknown graph embedded …
-
New Algorithm DYSCO Extracts Governing Equations from Latent Dynamics
Researchers have developed DYSCO, a novel multi-view temporal contrastive learning algorithm designed to identify latent dynamical systems and their governing equations from noisy, high-dimensional data. This method lev…
-
New complexity bound simplifies logconcave distribution sampling
Researchers have developed a new, unified complexity bound for sampling logconcave distributions. This bound is nearly tight and applies to various settings, including constrained and well-conditioned densities. The ana…
-
New JEPA Model Learns Sparse Representations with Rectified Distribution Matching
Researchers have developed Rectified LpJEPA, a novel approach to Joint-Embedding Predictive Architectures (JEPA) that aims to create more efficient and sparse representations. Unlike previous methods that favored dense …
-
AI privacy research finds no middle ground for hidden-state utility
A new research paper explores the challenge of maintaining privacy in AI models, specifically focusing on hidden-state privacy. The study found that out of 1,536 tested Gaussian release covariances for single-layer mode…
-
New Edgeworth Expansion Method Approximates Neural Network Output Deviations
Researchers have developed a method to approximate deviations in finite-width neural networks from their infinite-width Gaussian limits. This approach uses multidimensional Edgeworth expansions to quantify errors in Bay…
-
New algorithms tackle robust multiclass linear classification
Two new research papers published on arXiv introduce novel algorithms for multiclass linear classification under Gaussian distributions. The first paper focuses on achieving polynomial-time robust learning with dimensio…